Spatial Modeling of Wind Directions in South Africa

A spatial model is developed for wind directions in South Africa using data from one hundred stations. Wind directions at each station were modeled by mixtures of von Mises distributions, fitted by a Markov chain Monte Carlo algorithm. Spatial models were inferred for major mode, minor mode, concentration parameter estimate of major mode, concentration parameter estimate of minor mode, mean angle, mean resultant, circular variance, circular skewness and circular kurtosis.